Abstract: Techniques, systems, and articles of manufacture for determining statistics for cost-based optimization of a workflow. A method includes generating individual sets of statistics for each intermediate relation of a workflow, wherein said intermediate relations comprise results of stages of any plan of the workflow, and wherein each individual set of statistics computes cardinality of the corresponding intermediate relation; determining a global set of statistics for the workflow, wherein said global set of statistics comprises at least one of the individual sets of statistics for each of the intermediate relations; instrumenting a given plan of the workflow to collect the global set of statistics during execution; executing the given plan to collect the global set of statistics; and determining a plan of the workflow with the lowest cost by comparing the cost of multiple plans, wherein the cost of each plan is derived from the global set of statistics.

Abstract: Techniques, systems, and articles of manufacture for determining statistics for cost-based optimization of a workflow. A method includes generating individual sets of statistics for each intermediate relation of a workflow, wherein said intermediate relations comprise results of stages of any plan of the workflow, and wherein each individual set of statistics computes cardinality of the corresponding intermediate relation; determining a global set of statistics for the workflow, wherein said global set of statistics comprises at least one of the individual sets of statistics for each of the intermediate relations; instrumenting a given plan of the workflow to collect the global set of statistics during execution; executing the given plan to collect the global set of statistics; and determining a plan of the workflow with the lowest cost by comparing the cost of multiple plans, wherein the cost of each plan is derived from the global set of statistics.